In response to cellular stresses including heat, toxins, radiation, infection, inflammation, and oxidants, all cells produce a common set of heat shock proteins (Hsps) (Macario & de Macario 2000). Most heat shock proteins act as molecular chaperones. Chaperones bind and stabilize proteins at intermediate stages of folding and allow proteins to fold to their functional states. Hsp90 is the most abundant cytosolic Hsp under normal conditions. There are two human isoforms of Hsp90, a major inducible form Hsp90α and minor constitutively expressed form Hsp90β and two other closely related chaperones which are restricted in their intracellular location (Endoplasmic reticulum GP96/GRP94; mitochondrial form Hsp90β.TRAP1). The term HSP90 as used here includes all these analogues unless stated. Hsp90 binds proteins at a late stage of folding and is distinguished from other Hsps in that most of its protein substrates are involved in signal transduction. Hsp90 has a distinct ATP binding site, including a Bergerat fold characteristic of bacterial gyrase, topoisomerases and histidine kinases. It has been shown that ATP bound at the N-terminal pocket of Hsp90 is hydrolysed. This ATPase activity results in a conformational change in Hsp90 that is required to enable conformational changes in the client protein.
A dimerization domain and a second ATP binding site, which may regulate ATPase activity, is found near the c-terminus of Hsp90. Dimerization of HSP90 appears critical for ATP hydrolysis. Activation of Hsp90 is further regulated through interactions with a variety of other chaperone proteins and can be isolated in complex with other chaperones including Hsp70, Hip, Hop, p23, and p50cdc37. Many other co-chaperone proteins have also been demonstrated to bind HSP90. A simplified model has emerged in which ATP binding to the amino terminal pocket alters Hsp90 conformation to allow association with a multichaperone complex. First the client protein is bound to an Hsp70/Hsp40 complex. This complex then associates with Hsp90 via Hop. When ADP is replaced by ATP, the conformation of Hsp90 is altered, Hop and Hsp70 are released and a different set of co-chaperones is recruited including p50cdc37 and p23. ATP hydrolysis results in the release of these co-chaperones and the client protein from the mature complex. Ansamycin antibiotics herbimycin, geldanamycin (GA) and 17-allylamino-17-desmethoxygeldanamycin (17-AAG) are ATP binding site inhibitors that block the binding of ATP and prevent conversion to the mature complex (Grenert et. al., 1997. J Biol Chem., 272:23834-23850).
Despite Hsp90 being ubiquitously expressed, GA has a higher binding affinity for Hsp90 derived from tumour vs. normal cell lines (Kamal et. al., Nature 2003; 425: 407-410). GA also shows more potent cytotoxic activity in tumour cells and is sequestered at higher concentrations within tumours in xenograft mouse models (Brazidec J. Med. Chem. 2004, 47, 3865-3873). Furthermore the ATP-ase activity of Hsp90 is elevated in cancer cells and is an indication of the increased level of stress in these cells. Hsp90 gene amplification has also been reported to occur in the later stages of cancer (Jolly and Morimoto JNCI Vol. 92, No. 19, 1564-1572, 2000).
Increased genetic instability associated with the cancer phenotype leads to an increase in the production of non-native or mutant proteins. The ubiquitin pathway also serves to protect the cell from non-native or misfolded proteins, by targeting these proteins for proteasomal degradation. Mutant proteins are by their nature not native and therefore have the potential to show structural instability and an increased requirement for the chaperone system. (Giannini et al., Mol Cell Biol. 2004; 24(13):5667-76).
There is some evidence that Hsp90 is found primarily within “activated” multichaperone complexes in the tumour cells as opposed to “latent” complexes in normal cells. One component of the multichaperone complex is the cdc37 co-chaperone. Cdc37 binds Hsp90 at the base of the ATP binding site and could affect the off rates of inhibitors bound to Hsp90 in the “activated” state (Roe et. al., Cell 116, (2004), pp. 87-98). The client protein bound to the Hsp90-Hsp70 form of the chaperone complex is believed to be more susceptible to ubiquitination and targeting to the proteasome for degradation. E3 ubiquitin ligases have been identified with chaperone interacting motifs and one of these (CHIP) was shown to promote the ubiquitination and degradation of Hsp90 client proteins (Connell et al., 2001. Xu et al., 2002).
Hsp90 Client Proteins
The number of reported Hsp90 client proteins now exceeds 100. Since many of its client proteins are involved in cell signalling proliferation and survival, Hsp90 has received major interest as an oncology target. Two groups of client proteins, cell signalling protein kinases and transcription factors, in particular suggest Hsp90 regulation may have potential benefit as an anticancer therapy.
Hsp90 protein kinase client proteins implicated in cell proliferation and survival include the following:
c-Src
Cellular Src (c-Src) is a receptor tyrosine kinase, required for mitogenesis initiated by multiple growth factor receptors, including the receptors for epidermal growth factor receptor (EGFR), platelet-derived growth factor receptor (PDGFR), colony stimulating factor-1 (CSF-1R), and the basic fibroblast growth factor (bFGFR). C-Src is also overexpressed and activated in many of the same human carcinomas that overexpress EGFR and ErbB2. Src is also required for the maintenance of normal bone homeostasis through its regulation of osteoclast function.
b185erbB2
ErbB2 (Her2/neu) is a receptor tyrosine kinase overexpressed in a variety of malignancies including breast, ovarian, prostate, and gastric cancers. ErbB2 was originally identified as an oncogene and inhibition of Hsp90 results in the polyubiquitination and degradation of erbB2.
Polo Mitotic Kinase
Polo-like kinases (Plks) are important regulators of cell cycle progression during M-phase. Plks are involved in the assembly of the mitotic spindle apparatus and in the activation of CDK/cyclin complexes. Plk1 regulates tyrosine dephosphorylation of CDKs through phosphorylation and activation of Cdc25C. CDK1 activation in turn leads to spindle formation and entry into M phase.
Akt (PKB)
Akt is involved in pathways that regulate cell growth by stimulating cell proliferation and suppressing apoptosis. Hsp90 inhibition by ansamycins results in a reduction in the Akt half life through ubiquitination and proteasomal degradation. Binding of cdc37 to Hsp90 is also required for the down-regulation of Akt. Following ansamycin treatment cancer cells arrest in the G2/M phase of the cell cycle 24 hours after treatment and proceed to apoptosis 24-48 hours later. Normal cells also arrest 24 hours after ansamycin treatment, but do not proceed on to apoptosis.
c-Raf, B-RAF, Mek
The RAS-RAF-MEK-ERK-MAP kinase pathway mediates cellular responses to growth signals. RAS is mutated to an oncogenic form in approximately 15% of human cancers. The three RAF genes are serine/threonine kinases that are regulated by binding RAS.
EGFR
The epidermal growth factor receptor (EGFR) is implicated in cell growth, differentiation, proliferation, survival, apoptosis, and migration. Overexpression of EGFR has been found in many different cancers and activating mutations of its kinase domain appear to be pathogenic in a subset of adenocarcinoams of the lung.
Flt3
FMS-like tyrosine kinase 3 (FLT3) is a receptor tyrosine kinase involved in cell proliferation, differentiation and apoptosis. Flt3 activation also leads to the activation of phosphatidylinositol 3-kinase (PI3K) and RAS signal-transduction cascades.
c-Met
c-met is a receptor tyrosine kinase which binds hepatocyte growth factor (HGF) and regulates both cell motility and cell growth. c-met is overexpressed in tumours, including thyroid, stomach, pancreatic and colon cancer. HGF is also detected around the tumours, including liver metastases. This suggests that c-met and HGF play an important role in invasion and metastasis.
Cdk1, Cdk2, Cdk4, Cdk6
Cdk1, Cdk2, Cdk4, and Cdk6 drive the cell cycle. The activity of CDKs is regulated by their binding to specific subunits such as cyclins, inhibitory and assembly factors. The substrate specificity and timing of CDK activities is dictated by their interaction with specific cyclins. Cdk4/cyclin D and Cdk6/cyclin D are active in the G1 phase, Cdk2/cyclin E and Cdk2/cyclin A in S phase, and Cdc2/cyclin A and Cdc2/cyclin B in G2/M phase.
Cyclin-dependent kinase type 4 (CDK4), plays a key role in allowing cells to traverse G1 to S-phase transition of the cell cycle and is constitutively activated in many human cancers. The CDK4 activator, cyclin D1, is overexpressed and a CDK4 inhibitor, p16, is deleted in a variety of human tumours.
Cdk1/Cdk2 inhibitors have been developed which reversibly block normal cells in either the G1/S-phase or at the G2/M border. G2/M arrest is generally less well tolerated by the cells and consequently, they undergo apoptotic cell death. Since Hsp90 also is known to affect cell survival pathways this effect may be further amplified with an Hsp90 inhibitor.
Wee-1
The Wee-1 protein kinase carries out the inhibitory phosphorylation of CDC2 on tyrosine 15 (Tyr15). This is required for activation of the G2-phase checkpoint in response to DNA damage.
Hsp90 transcription factors implicated in cell proliferation and survival include the following:
Mutant p53
P53 is a tumour suppressor protein that causes cell cycle arrest and induces apoptosis. P53 is mutated in approximately half of all cancers. Mutant p53 associates with Hsp90 and is down-regulated in cancer lines treated with Hsp90 inhibitors, while wild type p53 levels were unaffected.
Progesterone Receptor/Estrogen Receptor/Androgen Receptor
Approximately 70% of post-menopausal women who develop breast cancer have tumours that express the estrogen receptor. The first line treatment of these patients is directed at preventing signalling through this pathway and thus inhibiting tumour growth. This can be done by ovarian ablation, treatment with gonadotrophin releasing hormone agonists, aromatase inhibition or treatment with specific agonists which bind to the estrogen receptor but prevent further signalling. Ultimately patients develop resistance to these interventions often as a consequence of crosstalk between the estrogen receptor and growth factor receptors located on the cell membrane. In the unliganded state estrogen receptors are complexed with Hsp90 which facilitates hormone binding.
Following binding to the mature receptor Hsp90 complex the liganded receptor can bind to hormone-response elements (HREs) within the regulatory regions of target genes involved in maintaining cell proliferation. Inhibition of Hsp90 initiates proteosomal degradation of the estrogen receptor thus preventing further growth signalling via this pathway. Prostate cancers are hormone-dependent malignancies that respond to therapeutic interventions which reduce circulating levels of testosterone or prevent testosterone binding to the androgen receptor. Although patients initially respond to these treatments most subsequently develop resistance via restoration of signalling via the androgen receptor. Prior to ligand binding the androgen receptor exists in a complex with Hsp90 and other co-chaperones including p23 and immunophilins. This interaction maintains the androgen receptor in a high-affinity ligand binding conformation. Inhibition of Hsp90 leads to proteosomal degradation of the androgen receptor and other co-chaperones which may sensitise the tumour to further hormonal therapies.
Mutated steroid hormone receptors that have arisen for example during anti-hormone therapy and which might be resistant to such therapies are likely to have a greater dependence on HSP90 for their stability and hormone binding function.
Hif-1a
Hypoxia inducible factor-1a (HIF-1a) is a transcription factor that controls the expression of genes which play a role in angiogenesis. HIF-1a is expressed in the majority of metastases and is known to associate with Hsp90. Ansamycin treatment of renal carcinoma cell lines leads to the ubiquitination and proteasomal degradation of HIF-1a.
Hsp90 inhibitors are capable of affecting a large number of targets significant to signal transduction in tumour cell proliferation. Signal transduction inhibitors which regulate the activities of a single target, may not be as efficacious due to signalling pathway redundancy and the rapid development of resistance.
By regulating multiple targets involved in cell signalling and cell proliferation HSP90 inhibitors may prove beneficial in the treatment of a wide spectrum of proliferative disorders.
ZAP70
ZAP-70, a member of the Syk-ZAP-70 protein tyrosine kinase family, is normally expressed in T cells and natural killer cells and has a critical role in the initiation of T-cell signaling. However, it is also expressed aberrantly in approximately 50% of cases of CLL, usually in those cases with unmutated B-cell receptor genes. The mutational status of immunoglobulin heavy-chain variable-region (IgVH) genes in the leukemic cells of chronic lymphocytic leukemia (CLL) is an important prognostic factor. The expression of ZAP-70 in CLL cells correlates with IgVH mutational status, disease progression, and survival. ZAP-70 positive CLL is more aggressive than ZAP-70 negative CLL indicating that ZAP-70 may be a key driver of malignancy in this disease. ZAP-70 is physically associated with HSP90 in B-CLL lymphoblasts thus the inhibition of Hsp90 may sensitise these cells to existing chemotherapy or monoclonal antibody therapy.
hERG
In the late 1990s a number of drugs, approved by the US FDA, had to be withdrawn from sale in the US when it was discovered they were implicated in deaths caused by heart malfunction. It was subsequently found that a side effect of these drugs was the development of arrhythmias caused by the blocking of hERG channels in heart cells. The hERG channel is one of a family of potassium ion channels the first member of which was identified in the late 1980s in a mutant Drosophila melanogaster fruitfly (see Jan, L. Y. and Jan, Y. N. (1990). A Superfamily of Ion Channels. Nature, 345(6277):672). The biophysical properties of the hERG potassium ion channel are described in Sanguinetti, M. C., Jiang, C., Curran, M. E., and Keating, M. T. (1995). A Mechanistic Link Between an Inherited and an Acquired Cardiac Arrhythmia: HERG encodes the lkr potassium channel. Cell, 81:299-307, and Trudeau, M. C., Warmke, J. W., Ganetzky, B., and Robertson, G. A. (1995). HERG, a Human Inward Rectifier in the Voltage-Gated Potassium Channel Family. Science, 269:92-95.
The elimination of hERG blocking activity remains an important consideration in the development of any new drug.
Heat Shock Proteins and Antitumour Drug Resistance
It has long been recognized that the native tertiary conformation of any given polypeptide is determined by its primary (amino acid) sequence. However, as explained above, it is now clear that the proper folding of many proteins in vivo requires the assistance of heat-shock proteins (Hsps) acting as molecular chaperones. While this chaperone function is important to normal cellular function under all conditions, it becomes crucial in cells which are stressed (for example by heat, hypoxia or acidosis).
Such conditions typically prevail in tumour cells, which exist in a hostile host environment. The upregulation of Hsps often seen in such cells is therefore likely to represent a mechanism by which malignant cells maintain the integrity of their proteomes under conditions which compromise protein folding. Thus, modulators or inhibitors of stress proteins in general (and Hsp90 in particular) represent a class of chemotherapeutics with the unique ability to inhibit multiple aberrant signaling pathways simultaneously. They can therefore exert antitumour effects whilst eliminating (or reducing the incidence of) resistance relative to other treatment paradigms.
Moreover, therapeutic anticancer interventions of all types necessarily increase the stresses imposed on the target tumour cells. In mitigating the deleterious effects of such stresses, Hsps are directly implicated in resisting the effects of cancer drugs and treatment regimens. Thus, modulators or inhibitors of stress protein function in general (and Hsp90 in particular) represent a class of chemotherapeutics with the potential for: (i) sensitizing malignant cells to anticancer drugs and/or treatments; (ii) alleviating or reducing the incidence of resistance to anticancer drugs and/or treatments; (iii) reversing resistance to anticancer drugs and/or treatments; (iv) potentiating the activity of anticancer drugs and/or treatments; (v) delaying or preventing the onset of resistance to anticancer drugs and/or treatments.
HSP90 Inhibitors and the Treatment of Hepatitis C and Other Viral Diseases
Infection of a host cell with viral RNA/DNA results in a substantial redirection of cellular protein synthesis towards key viral proteins encoded by the viral nucleic acid. The increased protein synthetic burden places a stress on the cell as a consequence of increased demand for energy and synthetic precursers. Upregulation of heat shock proteins is frequently a consequence of viral infection at least in part due to this stress. One function of the HSP induction may be to assist in the stabilization and folding of the high levels of ‘foreign’ protein generated in preparation for virus replication. In particular recent work has suggested that HSP90 is required for stable production of functional NS2/3 protease in Hepatitis C(HCV) replicon infected cells. HSP 90 inhibitors have also been demonstrated to block viral replication in in vitro systems. (Nagkagawa, S, Umehara T, Matsuda C, et al Biochem. Biophys. Res Commun. 353 (2007) 882-888; Waxman L, Witney, M et al PNAS 98 (2001) 13931-13935).
Glycogen Synthase Kinase
Glycogen Synthase Kinase-3 (GSK3) is a serine-threonine kinase that occurs as two ubiquitously expressed isoforms in humans (GSK3α & beta GSK3β). GSK3 has been implicated as having roles in embryonic development, protein synthesis, cell proliferation, cell differentiation, microtubule dynamics, cell motility and cellular apoptosis. As such GSK3 has been implicated in the progression of disease states such as diabetes, cancer, Alzheimer's disease, stroke, epilepsy, motor neuron disease and/or head trauma. Phylogenetically GSK3 is most closely related to the cyclin dependent kinases (CDKs).
The consensus peptide substrate sequence recognised by GSK3 is (Ser/Thr)-X-X-X-(pSer/pThr), where X is any amino acid (at positions (n+1), (n+2), (n+3)) and pSer and pThr are phospho-serine and phospho-threonine respectively (n+4). GSK3 phosphorylates the first serine, or threonine, at position (n). Phospho-serine, or phospho-threonine, at the (n+4) position appear necessary for priming GSK3 to give maximal substrate turnover. Phosphorylation of GSK3α at Ser21, or GSK3β at Ser9, leads to inhibition of GSK3. Mutagenesis and peptide competition studies have led to the model that the phosphorylated N-terminus of GSK3 is able to compete with phospho-peptide substrate (S/TXXXpS/pT) via an autoinhibitory mechanism. There are also data suggesting that GSK3α and GSKβ may be subtly regulated by phosphorylation of tyrosines 279 and 216 respectively. Mutation of these residues to a Phe caused a reduction in in vivo kinase activity. The X-ray crystallographic structure of GSK3β has helped to shed light on all aspects of GSK3 activation and regulation.
GSK3 forms part of the mammalian insulin response pathway and is able to phosphorylate, and thereby inactivate, glycogen synthase. Upregulation of glycogen synthase activity, and thereby glycogen synthesis, through inhibition of GSK3, has thus been considered a potential means of combating type II, or non-insulin-dependent diabetes mellitus (NIDDM): a condition in which body tissues become resistant to insulin stimulation. The cellular insulin response in liver, adipose, or muscle tissues, is triggered by insulin binding to an extracellular insulin receptor. This causes the phosphorylation, and subsequent recruitment to the plasma membrane, of the insulin receptor substrate (IRS) proteins. Further phosphorylation of the IRS proteins initiates recruitment of phosphoinositide-3 kinase (PI3K) to the plasma membrane where it is able to liberate the second messenger phosphatidylinosityl 3,4,5-trisphosphate (PIP3). This facilitates co-localisation of 3-phosphoinositide-dedependent protein kinase 1 (PDK1) and protein kinase B (PKB or Akt) to the membrane, where PDK1 activates PKB. PKB is able to phosphorylate, and thereby inhibit, GSK3α and/or GSKβ through phosphorylation of Ser9, or ser21, respectively. The inhibition of GSK3 then triggers upregulation of glycogen synthase activity. Therapeutic agents able to inhibit GSK3 may thus be able to induce cellular responses akin to those seen on insulin stimulation. A further in vivo substrate of GSK3 is the eukaryotic protein synthesis initiation factor 2B (eIF2B). eIF2B is inactivated via phosphorylation and is thus able to suppress protein biosynthesis. Inhibition of GSK3, e.g. by inactivation of the “mammalian target of rapamycin” protein (mTOR), can thus upregulate protein biosynthesis. Finally there is some evidence for regulation of GSK3 activity via the mitogen activated protein kinase (MAPK) pathway through phosphorylation of GSK3 by kinases such as mitogen activated protein kinase activated protein kinase 1 (MAPKAP-K1 or RSK). These data suggest that GSK3 activity may be modulated by mitogenic, insulin and/or amino acid stimulii.
It has also been shown that GSK3β is a key component in the vertebrate Wnt signalling pathway. This biochemical pathway has been shown to be critical for normal embryonic development and regulates cell proliferation in normal tissues. GSK3 becomes inhibited in response to Wnt stimulii. This can lead to the de-phosphorylation of GSK3 substrates such as Axin, the adenomatous polyposis coli (APC) gene product and β-catenin. Aberrant regulation of the Wnt pathway has been associated with many cancers. Mutations in APC, and/or β-catenin, are common in colorectal cancer and other tumours. β-catenin has also been shown to be of importance in cell adhesion. Thus GSK3 may also modulate cellular adhesion processes to some degree. Apart from the biochemical pathways already described there are also data implicating GSK3 in the regulation of cell division via phosphorylation of cyclin-D1, in the phosphorylation of transcription factors such as c-Jun, CCAAT/enhancer binding protein α (C/EBPα), c-Myc and/or other substrates such as Nuclear Factor of Activated T-cells (NFATc), Heat Shock Factor-1 (HSF-1) and the c-AMP response element binding protein (CREB). GSK3 also appears to play a role, albeit tissue specific, in regulating cellular apoptosis. The role of GSK3 in modulating cellular apoptosis, via a pro-apoptotic mechanism, may be of particular relevance to medical conditions in which neuronal apoptosis can occur. Examples of these are head trauma, stroke, epilepsy, Alzheimer's and motor neuron diseases, progressive supranuclear palsy, corticobasal degeneration, and Pick's disease. In vitro it has been shown that GSK3 is able to hyper-phosphorylate the microtubule associated protein Tau. Hyperphosphorylation of Tau disrupts its normal binding to microtubules and may also lead to the formation of intra-cellular Tau filaments. It is believed that the progressive accumulation of these filaments leads to eventual neuronal dysfunction and degeneration. Inhibition of Tau phosphorylation, through inhibition of GSK3, may thus provide a means of limiting and/or preventing neurodegenerative effects.
Cyclin Dependent Kinases
The process of eukaryotic cell division may be broadly divided into a series of sequential phases termed G1, S, G2 and M. Correct progression through the various phases of the cell cycle has been shown to be critically dependent upon the spatial and temporal regulation of a family of proteins known as cyclin dependent kinases (cdks) and a diverse set of their cognate protein partners termed cyclins. Cdks are cdc2 (also known as cdk1) homologous serine-threonine kinase proteins that are able to utilise ATP as a substrate in the phosphorylation of diverse polypeptides in a sequence dependent context. Cyclins are a family of proteins characterised by a homology region, containing approximately 100 amino acids, termed the “cyclin box” which is used in binding to, and defining selectivity for, specific cdk partner proteins.
Modulation of the expression levels, degradation rates, and activation levels of various cdks and cyclins throughout the cell cycle leads to the cyclical formation of a series of cdk/cyclin complexes, in which the cdks are enzymatically active. The formation of these complexes controls passage through discrete cell cycle checkpoints and thereby enables the process of cell division to continue. Failure to satisfy the pre-requisite biochemical criteria at a given cell cycle checkpoint, i.e. failure to form a required cdk/cyclin complex, can lead to cell cycle arrest and/or cellular apoptosis. Aberrant cellular proliferation, as manifested in cancer, can often be attributed to loss of correct cell cycle control. Inhibition of cdk enzymatic activity therefore provides a means by which abnormally dividing cells can have their division arrested and/or be killed. The diversity of cdks, and cdk complexes, and their critical roles in mediating the cell cycle, provides a broad spectrum of potential therapeutic targets selected on the basis of a defined biochemical rationale.
Progression from the G1 phase to the S phase of the cell cycle is primarily regulated by cdk2, cdk3, cdk4 and cdk6 via association with members of the D and E type cyclins. The D-type cyclins appear instrumental in enabling passage beyond the G1 restriction point, where as the cdk2/cyclin E complex is key to the transition from the G1 to S phase. Subsequent progression through S phase and entry into G2 is thought to require the cdk2/cyclin A complex. Both mitosis, and the G2 to M phase transition which triggers it, are regulated by complexes of cdk1 and the A and B type cyclins. During G1 phase Retinoblastoma protein (Rb), and related pocket proteins such as p130, are substrates for cdk(2, 4, & 6)/cyclin complexes. Progression through G1 is in part facilitated by hyperphosphorylation, and thus inactivation, of Rb and p130 by the cdk(4/6)/cyclin-D complexes. Hyperphosphorylation of Rb and p130 causes the release of transcription factors, such as E2F, and thus the expression of genes necessary for progression through G1 and for entry into S-phase, such as the gene for cyclin E. Expression of cyclin E facilitates formation of the cdk2/cyclin E complex which amplifies, or maintains, E2F levels via further phosphorylation of Rb. The cdk2/cyclin E complex also phosphorylates other proteins necessary for DNA replication, such as NPAT, which has been implicated in histone biosynthesis. G1 progression and the G1/S transition are also regulated via the mitogen stimulated Myc pathway, which feeds into the cdk2/cyclin E pathway. Cdk2 is also connected to the p53 mediated DNA damage response pathway via p53 regulation of p21 levels. p21 is a protein inhibitor of cdk2/cyclin E and is thus capable of blocking, or delaying, the G1/S transition. The cdk2/cyclin E complex may thus represent a point at which biochemical stimuli from the Rb, Myc and p53 pathways are to some degree integrated. Cdk2 and/or the cdk2/cyclin E complex therefore represent good targets for therapeutics designed at arresting, or recovering control of, the cell cycle in aberrantly dividing cells.
The exact role of cdk3 in the cell cycle is not clear. As yet no cognate cyclin partner has been identified, but a dominant negative form of cdk3 delayed cells in G1, thereby suggesting that cdk3 has a role in regulating the G1/S transition.
Although most cdks have been implicated in regulation of the cell cycle there is evidence that certain members of the cdk family are involved in other biochemical processes. This is exemplified by cdk5 which is necessary for correct neuronal development and which has also been implicated in the phosphorylation of several neuronal proteins such as Tau, NUDE-1, synapsin1, DARPP32 and the Munc18/Syntaxin1A complex. Neuronal cdk5 is conventionally activated by binding to the p35/p39 proteins. Cdk5 activity can, however, be deregulated by the binding of p25, a truncated version of p35. Conversion of p35 to p25, and subsequent deregulation of cdk5 activity, can be induced by ischemia, excitotoxicity, and β-amyloid peptide. Consequently p25 has been implicated in the pathogenesis of neurodegenerative diseases, such as Alzheimer's, and is therefore of interest as a target for therapeutics directed against these diseases.
Cdk7 is a nuclear protein that has cdc2 CAK activity and binds to cyclin H. Cdk7 has been identified as component of the TFIIH transcriptional complex which has RNA polymerase II C-terminal domain (CTD) activity. This has been associated with the regulation of HIV-1 transcription via a Tat-mediated biochemical pathway. Cdk8 binds cyclin C and has been implicated in the phosphorylation of the CTD of RNA polymerase II. Similarly the cdk9/cyclin-T1 complex (P-TEFb complex) has been implicated in elongation control of RNA polymerase II. PTEF-b is also required for activation of transcription of the HIV-1 genome by the viral transactivator Tat through its interaction with cyclin T1. Cdk7, cdk8, cdk9 and the P-TEFb complex are therefore potential targets for anti-viral therapeutics.
At a molecular level mediation of cdk/cyclin complex activity requires a series of stimulatory and inhibitory phosphorylation, or dephosphorylation, events. Cdk phosphorylation is performed by a group of cdk activating kinases (LAKs) and/or kinases such as wee1, Myt1 and Mik1. Dephosphorylation is performed by phosphatases such as cdc25(a & c), pp 2a, or KAP.
Cdk/cyclin complex activity may be further regulated by two families of endogenous cellular proteinaceous inhibitors: the Kip/Cip family, or the INK family. The INK proteins specifically bind cdk4 and cdk6. p16ink4 (also known as MTS1) is a potential tumour suppressor gene that is mutated, or deleted, in a large number of primary cancers. The Kip/Cip family contains proteins such as p21cip1,Waf1, p27Kip1 and p57kip2. As discussed previously p21 is induced by p53 and is able to inactivate the cdk2/cyclin(E/A) and cdk4/cyclin(D1/D2/D3) complexes. Atypically low levels of p27 expression have been observed in breast, colon and prostate cancers. Conversely over expression of cyclin E in solid tumours has been shown to correlate with poor patient prognosis. Over expression of cyclin D1 has been associated with oesophageal, breast, squamous, and non-small cell lung carcinomas.
The pivotal roles of cdks, and their associated proteins, in co-ordinating and driving the cell cycle in proliferating cells have been outlined above. Some of the biochemical pathways in which cdks play a key role have also been described. The development of monotherapies for the treatment of proliferative disorders, such as cancers, using therapeutics targeted generically at cdks, or at specific cdks, is therefore potentially highly desirable. Cdk inhibitors could conceivably also be used to treat other conditions such as viral infections, autoimmune diseases and neuro-degenerative diseases, amongst others. Cdk targeted therapeutics may also provide clinical benefits in the treatment of the previously described diseases when used in combination therapy with either existing, or new, therapeutic agents. Cdk targeted anticancer therapies could potentially have advantages over many current antitumour agents as they would not directly interact with DNA and should therefore reduce the risk of secondary tumour development.
Aurora Kinases
Relatively recently, a new family of serine/threonine kinases known as the Aurora kinases has been discovered that are involved in the G2 and M phases of the cell cycle, and which are important regulators of mitosis.
The precise role of Aurora kinases has yet to be elucidated but that they play a part in mitotic checkpoint control, chromosome dynamics and cytokinesis (Adams et al., Trends Cell Biol., 11: 49-54 (2001). Aurora kinases are located at the centrosomes of interphase cells, at the poles of the bipolar spindle and in the mid-body of the mitotic apparatus.
Three members of the Aurora kinase family have been found in mammals so far (E. A. Nigg, Nat. Rev. Mol. Cell. Biol. 2: 21-32, (2001)). These are:                Aurora A (also referred to in the literature as Aurora 2);        Aurora B (also referred to in the literature as Aurora 1); and        Aurora C (also referred to in the literature as Aurora 3).        
The Aurora kinases have highly homologous catalytic domains but differ considerably in their N-terminal portions (Katayama H, Brinkley W R, Sen S.; The Aurora kinases: role in cell transformation and tumorigenesis; Cancer Metastasis Rev. 2003 December; 22(4):451-64).
The substrates of the Aurora kinases A and B have been identified as including a kinesin-like motor protein, spindle apparatus proteins, histone H3 protein, kinetochore protein and the tumour suppressor protein p53.
Aurora A kinases are believed to be involved in spindle formation and become localised on the centrosome during the early G2 phase where they phosphorylate spindle-associated proteins (Prigent et al., Cell, 114: 531-535 (2003). Hirota et al, Cell, 114:585-598, (2003) found that cells depleted of Aurora A protein kinase were unable to enter mitosis. Furthermore, it has been found (Adams, 2001) that mutation or disruption of the Aurora A gene in various species leads to mitotic abnormalities, including centrosome separation and maturation defects, spindle aberrations and chromosome segregation defects.
The Aurora kinases are generally expressed at a low level in the majority of normal tissues, the exceptions being tissues with a high proportion of dividing cells such as the thymus and testis. However, elevated levels of Aurora kinases have been found in many human cancers (Giet et al., J. Cell. Sci. 112: 3591-361, (1999) and Katayama (2003). Furthermore, Aurora A kinase maps to the chromosome 20q13 region that has frequently been found to be amplified in many human cancers.
Thus, for example, significant Aurora A over-expression has been detected in human breast, ovarian and pancreatic cancers (see Zhou et al., Nat. Genet. 20: 189-193, (1998), Tanaka et al., Cancer Res., 59: 2041-2044, (1999) and Han et al., cancer Res., 62: 2890-2896, (2002).
Moreover, Isola, American Journal of Pathology 147, 905-911 (1995) has reported that amplification of the Aurora A locus (20ql3) correlates with poor prognosis for patients with node-negative breast cancer.
Amplification and/or over-expression of Aurora-A is observed in human bladder cancers and amplification of Aurora-A is associated with aneuploidy and aggressive clinical behaviour, see Sen et al., J. Natl. Cancer Inst, 94: 1320-1329 (2002).
Elevated expression of Aurora-A has been detected in over 50% of colorectal cancers, (see Bischoff et al., EMBO J., 17: 3052-3065, (1998) and Takahashi et al., Jpn. J. Cancer Res., 91: 1007-1014 (2000)) ovarian cancers (see Gritsko et al. Clin. Cancer Res., 9: 1420-1426 (2003), and gastric tumours Sakakura et al., British Journal of Cancer, 84: 824-831 (2001).
Tanaka et al. Cancer Research, 59: 2041-2044 (1999) found evidence of over-expression of Aurora A in 94% of invasive duct adenocarcinomas of the breast.
High levels of Aurora A kinase have also been found in renal, cervical, neuroblastoma, melanoma, lymphoma, pancreatic and prostate tumour cell lines Bischoff et al. (1998), EMBO J., 17: 3052-3065 (1998); Kimura et al. J. Biol. Chem., 274: 7334-7340 (1999); Zhou et al., Nature Genetics, 20: 189-193 (1998); Li et al., Clin Cancer Res. 9 (3): 991-7 (2003)].
Aurora-B is highly expressed in multiple human tumour cell lines, including leukemic cells [Katayama et al., Gene 244: 1-7)]. Levels of this enzyme increase as a function of Duke's stage in primary colorectal cancers [Katayama et al., J. Natl Cancer Inst., 91: 1160-1162 (1999)].
High levels of Aurora-3 (Aurora-C) have been detected in several tumour cell lines, even though this kinase tends to be restricted to germ cells in normal tissues (see Kimura et al. Journal of Biological Chemistry, 274: 7334-7340 (1999)). Over-expression of Aurora-3 in approximately 50% of colorectal cancers has also been reported in the article by Takahashi et al., Jpn J. Cancer Res. 91: 1007-1014 (2001)].
Other reports of the role of Aurora kinases in proliferative disorders may be found in Bischoff et al., Trends in Cell Biology 9: 454-459 (1999); Giet et al. Journal of Cell Science, 112: 3591-3601 (1999) and Dutertre, et al. Oncogene, 21: 6175-6183 (2002).
Royce et al report that the expression of the Aurora 2 gene (known as STK15 or BTAK) has been noted in approximately one-fourth of primary breast tumours. (Royce M E, Xia W, Sahin A A, Katayama H, Johnston D A, Hortobagyi G, Sen S, Hung M C; STK15/Aurora-A expression in primary breast tumours is correlated with nuclear grade but not with prognosis; Cancer. 2004 Jan. 1; 100(1):12-9).
Endometrial carcinoma (EC) comprises at least two types of cancer: endometrioid carcinomas (EECs) are estrogen-related tumours, which are frequently euploid and have a good prognosis. Nonendometrioid carcinomas (NEECs; serous and clear cell forms) are not estrogen related, are frequently aneuploid, and are clinically aggressive. It has also been found that Aurora was amplified in 55.5% of NEECs but not in any EECs (P< or =0.001) (Moreno-Bueno G, Sanchez-Estevez C, Cassia R, Rodriguez-Perales S, Diaz-Uriarte R, Dominguez O, Hardisson D, Andujar M, Prat J, Matias-Guiu X, Cigudosa J C, Palacios J. Cancer Res. 2003 Sep. 15; 63(18):5697-702).
Reichardt et al (Oncol Rep. 2003 September-October; 10(S):1275-9) have reported that quantitative DNA analysis by PCR to search for Aurora amplification in gliomas revealed that five out of 16 tumours (31%) of different WHO grade (1× grade II, 1× grade III, 3× grade IV) showed DNA amplification of the Aurora 2 gene. It was hypothesized that amplification of the Aurora 2 gene may be a non-random genetic alteration in human gliomas playing a role in the genetic pathways of tumourigenesis.
Results by Hamada et al (Br. J. Haematol. 2003 May; 121(3):439-47) also suggest that Aurora 2 is an effective candidate to indicate not only disease activity but also tumourigenesis of non-Hodgkin's lymphoma. Retardation of tumour cell growth resulting from the restriction of this gene's functions could be a therapeutic approach for non-Hodgkin's lymphoma.
In a study by Gritsko et al (Clin Cancer Res. 2003 April; 9(4):1420-6)), the kinase activity and protein levels of Aurora A were examined in 92 patients with primary ovarian tumours. In vitro kinase analyses revealed elevated Aurora A kinase activity in 44 cases (48%). Increased Aurora A protein levels were detected in 52 (57%) specimens. High protein levels of Aurora A correlated well with elevated kinase activity.
Results obtained by Li et al (Clin. Cancer Res. 2003 March; 9(3):991-7) showed that the Aurora A gene is overexpressed in pancreatic tumours and carcinoma cell lines and suggest that overexpression of Aurora A may play a role in pancreatic carcinogenesis.
Similarly, it has been shown that Aurora A gene amplification and associated increased expression of the mitotic kinase it encodes are associated with aneuploidy and aggressive clinical behaviour in human bladder cancer. (J. Natl. Cancer Inst. 2002 Sep. 4; 94(17):1320-9).
Investigation by several groups (Dutertre S, Prigent C., Aurora-A overexpression leads to override of the microtubule-kinetochore attachment checkpoint; Mol. Interv. 2003 May; 3(3):127-30 and Anand S, Penrhyn-Lowe S, Venkitaraman A R., Aurora-A amplification overrides the mitotic spindle assembly checkpoint, inducing resistance to Taxol, Cancer Cell. 2003 January; 3(1):51-62) suggests that overexpression of Aurora kinase activity is associated with resistance to some current cancer therapies. For example overexpression of Aurora A in mouse embryo fibroblasts can reduce the sensitivity of these cells to the cytotoxic effects of taxane derivatives. Therefore Aurora kinase inhibitors may find particular use in patients who have developed resistance to existing therapies.
On the basis of work carried out to date, it is apparent that inhibition of Aurora kinases, particularly Aurora kinase A and Aurora kinase B, will prove an effective means of arresting tumour development.
Harrington et al (Nat Med. 2004 March; 10(3):262-7) have demonstrated that an inhibitor of the Aurora kinases suppresses tumour growth and induces tumour regression in vivo. In the study, the Aurora kinase inhibitor blocked cancer cell proliferation, and also triggered cell death in a range of cancer cell lines including leukaemic, colorectal and breast cell lines. In addition, it has shown potential for the treatment of leukemia by inducing apoptosis in leukemia cells. VX-680 potently killed treatment-refractory primary Acute Myelogenous Leukemia (AML) cells from patients (Andrews, Oncogene, 2005, 24, 5005-5015).
Recent reports indicate that Aurora kinases A and B are overexpressed in human leukaemia cells and that a small molecule Aurora kinase inhibitor is active against the growth of primary acute myeloid cells in vitro (Harrington et al, 2004). Moreover it has recently been reported that the product of the PML gene that is disrupted in acute promyelocytic leukaemia by a t(15:17) translocation (PML3), interacts with Aurora A and suppresses its kinase activity. Further evidence is emerging that PML is a tumor suppressor and that its disruption is not limited to leukaemias but may also be common in lymphomas and some solid tumors (Xu et al, Molecular Cell 17: 721-732, 2005).
Cancers which may be particularly amenable to Aurora inhibitors include breast, bladder, colorectal, pancreatic, ovarian, non-Hodgkin's lymphoma, gliomas and nonendometrioid endometrial carcinomas. Leukemias particularly amenable to Aurora inhibitors include Acute Myelogenous Leukemia (AML), chronic myelogenous leukaemia (CML), B-cell lymphoma (Mantle cell), and Acute Lymphoblastic Leukemia (ALL). Further leukemias include acute promyelocytic leukaemia.
Overexpression of Aurora kinase A has been identified as an independent predictor of poor prognosis in patients with medulloblastoma, a highly malignant primitive neuroectodermal tumor of the cerebellum (Neben et al., Cancer Research, 64: 3103-3111 (2004).
Ancillary Compounds
A wide variety of compounds of the formula (I′) find application in the combinations of the invention, as described in detail below.
The compounds of formula (I′) for use in the combinations of the invention therefore include the compound classes (a) and (b) as described herein, so including the compounds of WO 2005/002552 and WO 2006/070195 corresponding to those of formula (I′) described therein and sub-groups, embodiments and examples thereof (as also therein defined). The content of PCT/GB2004/002824 (WO 2005/002552) describing the various subgroups, embodiments and examples of compounds of formula (I′) are hereby incorporated herein by reference, as are the compounds of the formula (I′) described in WO 2006/070195 (the contents of which are also incorporated herein by reference).
Auxiliary Compounds
A wide variety of optional auxiliary compounds may be further combined with the combinations of the invention, as described in detail below. The optional auxiliary compounds may be anti-cancer agents.
WO 99/29705 (Glycomed et al) disclose a class of glycomimetic compounds having a number of possible uses including the treatment of cancer. One compound specifically disclosed in WO 99/29705 is the compound 2-(2-hydroxy-benzoyl)-1,2,3,4-tetrahydro-isoquinoline-3-carboxylic acid.
Our earlier International application PCT/GB2006/001382 discloses hydroxybenzoic acid amides as Hsp90 inhibitors.